You are on page 1of 37

TECHNICAL REPORT AD ________________

NATICK/TR-09/021L

PHOTOSIMULATION CAMOUFLAGE DETECTION TEST

by
(b) (6)
(b) (6)
(b) (6)(b) (6)
(b) (6)(b) (6)
(b) (6)
(b) (6)
(b) (6)

*Science Applications International Corporation (SAIC)


Natick, MA 01760

June 2009

Final Report
March 2007 – March 2009

Distribution authorized to U.S. Government Agencies and their contractors; specific


authority, March 2007. Other requests for this document shall be referred to Commander,
U.S. Army Natick Soldier Research, Development and Engineering Center (NSRDEC),
Kansas Street, ATTN: RDNS-WSH-C ((b) (6)(b) (6) Natick, MA 01760-5020.
(b) (4)
(b) (4)
(b) (4)
(b) (4)

U.S. Army Natick Soldier Research, Development and Engineering Center


Natick, Massachusetts 01760-5020
Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including
suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway,
Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of
information if it does not display a currently valid OMB control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To)
23-06-2009 Final March 2007 – March 2009
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER

PHOTOSIMULATION CAMOUFLAGE DETECTION TEST


5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER


63001
6. AUTHOR(S) 5d. PROJECT NUMBER

(b) (6)(b) (6)(b) (6)(b) (6)(b) (6)(b) (6)(b) (6)


(b) (6)(b) (6)(b) (6) 5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT


U.S. Army Natick Soldier Research, Development and Engineering Center NUMBER
ATTN: RDNS-WSH-C
NATICK/TR-09-021L
Kansas St., Natick, MA 01760-5020
9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

11. SPONSOR/MONITOR’S REPORT NUMBER(S)

12. DISTRIBUTION / AVAILABILITY STATEMENT


Distribution authorized to U.S. Government Agencies and their contractors; specific authority, March 2007. Other
requests for this document shall be referred to Commander, U.S. Army Natick Soldier Research, Development and
Engineering Center (NSRDEC), Kansas Street, ATTN: RDNS-WSH-C (b) (6)(b) (6) Natick, MA 01760-5020.
13. SUPPLEMENTARY NOTES
*Science Applications International Corporation (SAIC), 190 N. Main St., Natick, MA 01760
14. ABSTRACT
This report documents the methodology used and the results of a computerized photosimulation evaluation to quantify
the military effectiveness of various camouflage patterns across different environments. The evaluation was conducted
by the U.S. Army Natick Soldier Research, Development and Engineering Center (NSRDEC) between March 2007 and
March 2009. This study extended a digital blending evaluation conducted by NSRDEC in 2006 of two different
camouflage patterns in a given environment to multiple environments. The photosimulation evaluation technique used
was developed by NSRDEC in collaboration with the Aberdeen Test Center in 2008. The objective was to evaluate the
detection performance of camouflage patterns by obtaining visual R50 values (range for 50% probability of detection) for
a total of 18 standard, foreign and experimental camouflage patterns in various terrain environments under daylight
conditions. Probability of detection data were collected from 913 observers. The data clearly show that environment-
specific patterns provide the best camouflage, i.e., the lowest probability of detection, in their respective environments.
Recommendations for further analysis of the data are included in the report.
15. SUBJECT TERMS
DYES PATTERNS CAMOUFLAGE ARMY PERSONNEL
COLORS CONTRAST REFLECTANCE TARGET DETECTION
FABRICS BLENDING DESIGN CRITERIA INFRARED DETECTION
TERRAIN DETECTION COMBAT UNIFORM SIGNATURE REDUCTION
UNIFORMS SIGNATURES VISUAL PERCEPTION PATTERN RECOGNITION
CAMOUFLAGE PATTERNS ENVIRONMENTAL CONDITIONS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE ABSTRACT OF PAGES (b) (6)(b) (6)(b) (6)(b) (6)
U U U SAR 36 19b. TELEPHONE NUMBER (include area code)
508-233-4150/5427
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std. Z39.18
(This page has intentionally been left blank.)
Table of Contents

List of Figures ………………………………………………….……………………………… iv

List of Tables …………………………………………………….……………………………. v

Preface …………………………………………………….……..………………..……………. vi

Acknowledgements ……………………………………….………………….……………….. vii

1.0 Introduction ………………………………………….…………………………………… 1

2.0 Methodology ……...…………………………..……….…………………………………… 1

2.1 Imagery Collection …………………………………..…….……….………………. 1


2.2 Observer Data Collection ……………………………….………………………….. 4
2.3 Observers …………………………………………...….…………………………… 6
2.4 Patterns ………………………………………………..…………………………….. 6

3.0 Data Analysis ……………………………………….…….……………………………… 10

3.1 Data Adjustments ………………………….….……..……………………………... 11

4.0 Results ………………………………………….….……….………………………………. 12

4.1 Medians by Rule ……………………………………….……………………………. 12


4.2 Medians by Pattern / Environment ……………………………….……………….. 13
4.3 Ranking …………………………………………………………..…………………. 15
4.4 Results by Pattern ……………………………….……………..…………………… 16
4.4.1 UCP ..……………………………………………..………………………... 16
4.4.2 Desert Brush ………………….……………….…………………………… 16
4.4.3 Desert MARPAT …………….………….…….……………………………. 17
4.4.4 MultiCam® …………………….………...…….…………………………… 17
4.4.5 Syria ……………………….………….……….……………………………. 18
4.4.6 Performance …….………………………..………………………………… 18

5.0 Discussion ………………………………….……………………………………………… 20

6.0 Conclusions and Recommendations …………………….……………………………….. 21

7.0 References ……………………..………...……………..………………………………….. 23

Appendix A. Observer Instructions ……………..………..….…..………………………..… 25

Appendix B. Random Orders for Each Pattern ...……………………..…………………… 27

iii
List of Figures

Figure 1 Desert Scenes ……..……………………………………………………………. 3

Figure 2 Woodland Scenes ……..……………….……………………………………….. 3

Figure 3 Urban Scenes ……..………………….…………………………………………. 3

Figure 4 Test Set-up ………………………………………………………………………. 5

Figure 5 Scoring Box ……………………………………………………………………… 6

Figure 6 Universal Camouflage Pattern ……………..….……………………………… 7

Figure 7 Woodland Patterns …………………………………………………………….. 8

Figure 8 Desert Patterns …………………………………………………………………. 8

Figure 9 Commercial Patterns …………………………………………………………… 9

Figure 10 Patterns Eliminated from Final Analysis ………….………..………………… 9

Figure 11 Sample Observer Data ………………………….………..…………………….. 11

Figure 12 Medians by Rule …….…………………………………….…………………….. 12

Figure 13 Pattern R50 by Environment ……………………...………………………….… 13

Figure 14 Range of Data for Nine Scenes ……...……………..…………………………... 14

Figure 15 Differences in Median Detection Distances Relative to the UCP ……….… 19

iv
List of Tables

Table 1 Imagery Collection Locations and Dates ……….…….………………………. 2

Table 2 Environments and Scenes ………………………….…………………………… 2

Table 3 Observer Demographics ……………………………………….……………….. 6

Table 4 Number of Observers Evaluating Each Pattern by Scene ……….…………... 10

Table 5 Detection Analysis Rules ……..………………………..……………………….. 10

Table 6 Range of Medians by Environment …….…………..…………….…………… 14

Table 7 Ranking of Patterns by Scene ………….………………………………………. 15

Table 8 Ranking of Patterns by Environment ……..…………………………………. 15

Table 9 Statistically Significant Results Relative to the UCP …………………………. 16

Table 10 Statistically Significant Results Relative to the Desert Brush ……………….. 17

Table 11 Statistically Significant Results Relative to the Desert MARPAT …………… 17

Table 12 Statistically Significant Results Relative to the MultiCam® ………..………. 18

Table 13 Statistically Significant Results Relative to the Syria ……..…………………. 18

v
Preface

This report documents the methodology used and the results of a computerized photosimulation
evaluation to quantify the military effectiveness of various camouflage patterns across different
environments. The evaluation was conducted by the U.S. Army Natick Soldier Research, Development
and Engineering Center (NSRDEC) during the period March 2007 – March 2009 under program element
number 63001. The objective was to evaluate the detection performance of camouflage patterns by
obtaining visual R50 values (range for 50% probability of detection) for a total of 18 standard, foreign and
experimental camouflage patterns in various terrain environments under daylight conditions. Probability
of detection data were collected from 913 observers.

vi
Acknowledgements

The authors would like to acknowledge the support they received from the Aberdeen Test Center during
all phases of execution of this effort: (b) (6) supported imagery
collection at Ft. Campbell and Devens, and (b) (6) programmed the Photosimulation Test.
Observer data collection was supported by personnel from the U.S. Army Natick Soldier Research
Development Engineering Center (NSRDEC): (b) (6)
(b) (6) and(b) (6) from Ft. Benning’s Test & Evaluation, Soldier
Requirements Division allowed us to collect data in conjunction with their Post Combat Survey at Ft.
Drum and Ft. Lewis.

We would like to thank the following units for their support with the Photosimulation Observer test:

Mountain Warfare Training Center, VT:


102 Infantry National Guard, CT
Mountain Warfare School students from various units through-out the US

Ft Bliss, TX:
Warrior Transition Unit
6th Brigade Air Defense Artillery

Ft Drum, NY:
3-10 General Support Aviation Battalion
277 Aviation Support Battalion
HHC 10th Combat Aviation Brigade
2-10 Aviation

Ft Lewis, WA:
4-9 Infantry
2-23 Infantry
1-38 Infantry
2-1 Cavalry
2-12 Field Artillery

vii
(This page has intentionally been left blank.)

viii
PHOTOSIMULATION CAMOUFLAGE DETECTION TEST

1.0 Introduction
In 2006 the U.S. Army Natick Soldier Research, Development and Engineering Center (NSRDEC)
conducted a digital blending evaluation of two different camouflage patterns: Universal Camouflage
Pattern (UCP) and the MultiCam® Pattern1. The results, based on 12,000 data points, indicated that
MultiCam® performed significantly better than the UCP in most conditions. Though this evaluation
determined which pattern blends better in a given environment, further research was necessary to
determine the military effectiveness of various camouflage patterns across environments. In 2008,
NSRDEC worked with the Aberdeen Test Center to develop a computerized photosimulation evaluation
to quantify the effectiveness of various camouflage patterns. The objective was to evaluate the detection
performance of camouflage patterns by obtaining visual R50 values (range for 50% probability of
detection) for a total of 18 standard, foreign and experimental camouflage patterns in various terrain
environments under daylight conditions. Probability of detection data were collected from 913 observers.
This report documents the methodology used and the results of this evaluation.

2.0 Methodology
Visual camouflage evaluations typically require military observers to participate in multi-day field
evaluations. With this approach, there is little control over the environmental conditions (e.g., lighting,
shadows, weather), test conditions (e.g., background variation, noise) and observer pool. To better
control these factors, the North Atlantic Treaty Organization, Research and Technology Organization
(NATO RTO) Task group-SCI-095 developed guidance for photosimulation data collection and
analysis2,3. This methodology has the advantage of capturing performance under more controlled
environmental conditions, as well as bringing field data to the observers rather than bringing the observers
to the field. It also has the combined benefits of reducing the costs of the evaluations and increasing the
statistical confidence in the results by including more observers. These NATO RTO guidelines were
generally followed during the imagery data collection, observer data collection and data analysis portions
of this effort. Additional approaches were also employed and are described below.

2.1 Imagery Collection


Imagery was collected in three environments of interest to the US Army: desert, urban and woodland. The
desert imagery was collected near the MOUT site at Ft. Irwin, CA., former Ft. Devens, MA was used for
the woodland imagery, and the Cassidy MOUT site at Ft. Campbell, KY was used for the urban imagery
(Table 1). Three different scenes were used in each environment, as shown in Figures 1 – 3. Images in
these figures were taken at 400 m and the arrows indicate the target location in each scene. The three
scenes in each environment were selected in an effort to vary either lighting conditions on the target or
clutter surrounding the target. However, these scenes represent only a small fraction of the variation that
may be encountered in current or future operations.

Imagery was collected at 11 distances for each desert and woodland scene and at 7 distances for each
urban scene (Table 2). Due to limitations in the areas surrounding target locations, 400 m was the
maximum distance that could be obtained at the urban and woodland sites. This limitation dictated the

1
maximum distance used at the desert site. The minimum imagery distance was 25 m, except in the urban
environment where the targets were clearly visible at 125 m due to the nature of that environment. While
the goal at each site was to have all patterns undetectable at the longest distance in accordance with the
NATO guidelines, this was not always possible, especially for environment–specific patterns used in the
“wrong” environment (e.g., a woodland pattern in a desert environment).

Distances were measured using either a construction grade tape measure or a laser range finder. Imagery
capture locations were measured and marked prior to imagery collection, which began at approximately
1000 hrs and ended at approximately 1300 hrs. This provided a worst case scenario in terms of pattern
effectiveness in bright sunlight. One scene was completed per day, to minimize change in illumination
conditions. Imagery data collection for woodland was conducted on two separate occasions, as several
patterns (i.e., Desert British, Bulldog, France, and Woodland British) were not available during the first
woodland imagery collection. All efforts were made to collect the second set under the same
environmental and lighting conditions as were present during the first set.

All imagery was shot using a Nikon D200 digital camera. A focal length of 35 mm was used in order to
approximate the resolution that a human observer in the field would see at the same range. All images
were taken at three exposure levels (normal exposure, one level up, and one level down). The images
from the camera were 2896 pixels high by 1944 pixels wide and decreased slightly to 2400 pixels by
1607 pixels for use in the photosimulation evaluation.

Imagery Collection Locations and Dates


Table 1
Environment Location Dates
7 –10 March 2007
Woodland Devens, MA
15 – 17 July 2008
Desert Ft Irwin, CA 30 Sept – 5 Oct 2007
Urban Ft Campbell, KY 17 – 20 March 2008

Environments and Scenes


Table 2
Environment Scenes Distance (m)
25, 50, 75, 100, 125,
Woodland Woodland Woodland Woodland
150, 175, 200, 250,
(Devens, MA) Shade Partial Shade Bright Sun
300, 400
Urban Small 125, 150, 175, 200,
Urban Urban Door Urban Corner
Building 250, 300, 400
(Fort Campbell, KY)
25, 50, 75, 100, 125,
Desert Desert Rock Desert Sage Desert Sand 150, 175, 200, 250,
(Fort Irwin, CA) 300, 400

2
Sand and Rock Sage
Desert Scenes
Figure 1

Bright Sun Partial Shade Shade


Woodland Scenes
Figure 2

Corner, Small Building and Door


Urban Scenes
Figure 3

3
All imagery was taken with the target in the center of the image. Furthermore, the target stood facing
away from the camera to maximize the amount of pattern visible to the observers and to eliminate any
facial cues. The target’s standing location was marked to ensure each pattern was evaluated against the
exact same background. At a given distance, personnel donning the uniforms quickly rotated through the
18 uniforms. The camera was then moved to the next closest distance, and the process was repeated until
the imagery of all uniforms at all distances was captured. Because target detection was the focus of this
evaluation, full targets were in the field of view. That is, in the urban scenes, targets were not hidden in
windows or doorways, and in the woodland scenes they were not hidden behind brush. Hiding the targets
would have decreased the number pixels on a target, and pattern detection would have been confounded.
All targets used were approximately the same physical size.

2.2 Observer Data Collection


Prior to the data collection, the observer was given both written and oral instructions followed by a
practice set of imagery. Written instructions are included in Appendix A. The imagery used in the
practice set consisted of imagery collected at a location not used in the actual photosimulation evaluation.
All observer questions were answered before the test began, and discussion between observers was not
allowed during the evaluation.

The observer data collection was conducted using four individual stations. Each station consisted of a
laptop, mouse and calibrated monitor. The evaluation ran from a laptop that was not accessible to the
observer. Observers used only the mouse and monitor. The monitor used was a Samsung SyncMaster
214T, 21.3-inch TFT-PVA. It has a response time of 8ms (G to G) 900:1 contrast ratio, a brightness of
300 cd/m2, a pixel pitch of 0.270mm, a 1600 x 1200 maximum resolution, a scanning frequency of 30-81
kHz horizontal and 56-75 Hz vertical, and a horizontal/vertical viewing angle of 178/178. Figure 4 shows
the typical experimental set-up where the viewing distance was set to 36” to match the magnification of
the unaided eye in the field4. The calculation for the viewing distance is presented below. This distance
was marked on the table with tape, and observers were instructed not to lean in closer than the mark.
Each observer was supervised by a test controller (i.e., one test controller for two observers) to ensure
proper procedures were followed. Test controllers were situated behind the observer to prevent
interference during the photosimulation evaluation.

Viewing eye distance calculation:


EDlab = Rfield x HOMlab
AH field
where:
EDlab = Eye distance in the lab
Rfield = Range in the field
HOMlab = Height of monitor
AHfield = Actual target height in the field

4
Test Set-up
Figure 4
The program for the photosimulation evaluation was developed by Aberdeen Test Center using Adobe
Director, version 8.5. Each run, which was one pattern evaluated over all nine scenes, was set up prior to
the observer data collection efforts. To set up a run the test controller programmed the pattern,
environment, scene and sector of the image to display (i.e., left, right or center). Although the images
were captured with the target in the center, the software allowed the target to be displayed on the left,
right or center of the monitor. The targets, as seen by the observers, were always in the same sector (left,
right or center) within a series. Presentation of environment, scene and sector was randomized within
runs, and two random order presentations were used for each pattern. See Appendix B for the random
orders used for each pattern.

Each observer evaluated one pattern over all nine scenes (also referred to as one run). For a given scene,
the series of images started at 400 m and moved incrementally closer to the target. The observers were
instructed to click on the target in each image throughout the entire series, if and when the target was
detected. Upon detection, a “next” button could be used to advance to the next closest image in the series;
otherwise, the image automatically advanced to the next closest image after 14 s. All images in a series
were viewed, even after initial detection. The software allowed for real-time scoring and recorded, in an
Excel file, both time and location of the observer’s click on the target. Figure 5 shows an example of the
scoring box used, which was one and a half times the height and width of the target. A click inside the
scoring box was recorded as a “hit”. A click outside the scoring box was recorded as a ”miss”. When the
observer did not click anywhere on the image, “no response” was recorded. Observers were allowed to
click unlimited times on an image; however, only the first click (initial detection) was used as the
detection data point.

5
Scoring Box
Figure 5

2.3 Observers

A total of 885 observers were included in the final analysis. All observers were active duty Army
Soldiers, National Guard troops or retired military with theatre experience within the last 5 years. They
were drawn from various populations: NSRDEC; Mountain Warfare Training School, VT; Ft. Drum, NY;
Ft. Bliss, TX and Ft. Lewis, WA. Table 3 summarizes observer demographic information. Although
data were collected from 913 observers, some data were dropped from final analysis due to: incomplete
runs, observers having difficulty completing the evaluations or computer and/or equipment issues.

Observer Demographics
Table 3
Location Number (N) Mean Age Male Female
NSRDEC 26 40 24 2
Mountain Warfare
47 27 47 0
Training Center
Ft Bliss 219 33 189 30
Ft Drum 241 30 209 32
Ft Lewis 352 26 352 0
Total 885 30 821 64

2.4 Patterns
Eighteen patterns were included in the imagery collection and observer data collection (Figures 6-10).
Guidance for pattern selection was given by BG Brown, PEO-Soldier in March 2007. Guidance included
patterns of foreign countries, patterns presently available on the commercial market and the UCP. Final
pattern selection was controlled by pattern availability. The Woodland and Desert Battle Dress Uniforms
(BDU) were not included in this evaluation because, at the time of pattern selection, they were not being
worn by the U.S. Army.

6
All targets wore desert combat boots and a green Advanced Combat Helmet (ACH). Helmet covers were
not worn because of the lack of availability of covers for most patterns. Deployment and training
schedules limited the availability of observers during the short data collection period. Therefore, to focus
data collection on the patterns of greatest interest, the patterns were prioritized into two tiers. When a
complete data set was collected on the Tier 1 patterns, which are annotated in Figures 6-10 with an
asterisk, observer data collection then began on the Tier 2 patterns.

Although detection data were collected on 18 patterns, five patterns (i.e., Sweden, Spec4 Woodland,
Spec4 Urban, North Korea, and Woodland British) were eliminated from the final data analysis. The two
Spec4 patterns were not available for desert image collection; therefore, they did not have a complete data
set. Sweden, North Korea and Woodland British were eliminated, due to being the worst performers in
two out of the three environments. Their similarity to other woodland patterns was further justification,
although it must be noted that detection data are available for further analysis, if desired.

Universal Camouflage
Pattern*
Figure 6

7
China France Woodland MARPAT* Iraq
Woodland Patterns
Figure 7

Desert Brush* Desert MARPAT* Desert British Syria*


Desert Patterns
Figure 8

8
Natural Gear* Bulldog (Mirage) * Mossy Oak* MultiCam® *
Commercial Patterns
Figure 9

Sweden Spec4 Woodland Spec4 Urban North Korea Woodland British


Patterns Eliminated from Final Analysis
Figure 10

Table 4 shows the number of observers for each pattern and scene. There were approximately 40
observers for each pattern, with three exceptions. Bulldog has fewer data points in the woodland
environment because woodland imagery was not available at the onset of the evaluation. MultiCam® and
UCP had over 70 responses in most conditions because, initially, these two patterns were the primary
patterns of interest. Therefore, the first observer data collection location (Mountain Warfare Training
Center) collected data for only the MultiCam® and UCP patterns.

9
Number of Observers Evaluating Each Pattern by Scene
Table 4
Desert Urban Woodland
Bright Partial
Rock Sand Sage Door Building Corner Shade
Sun Sun
Desert Brush 48 48 48 48 48 48 48 48 48
Syria 45 38 45 45 45 45 45 45 45
Desert British 49 49 49 49 49 49 46 46 46
Desert
45 45 45 45 45 45 45 45 45
MARPAT
MultiCam® 76 77 77 63 76 76 76 77 77
China 47 47 47 47 47 47 47 47 47
Woodland
45 45 45 45 45 45 45 45 45
MARPAT
Natural Gear 50 50 50 50 45 50 50 50 50
Mossy Oak 43 38 43 43 43 43 43 43 43
Bulldog 45 45 45 45 45 45 29 29 29
Iraq 44 44 44 44 44 44 44 44 44
France 45 45 45 45 45 45 43 43 43
UCP 70 70 70 63 70 70 70 70 70

3.0 Data Analysis


Statistical Package for the Social Sciences (SPSS)5 was used to calculate the median R50 values using
analysis techniques outlined in the NATO guidelines. The initial analysis found unusual patterns in
observer responses for approximately 20% of the results. In some cases, there were “hits” at a far
distance, followed by one or more closer distances where there were “misses” or “no response”. Figure
11 illustrates examples of response data showing hits, misses and no responses. Green blocks represent
hits at a given distance, while the misses and no responses are indicated by the red and yellow blocks,
respectively. During observer data collection, some of the computers appeared to run slower than others.
It was unclear as to whether data were missing because of computer issues or whether observers were
making lucky guesses at the longer distances. In order to address this issue, two alternative detection
rules were developed (see Table 5).

Detection Analysis Rules


Table 5
Move from 25m towards 400m until you reach the maximum distance
Conservative (C)
without a MISS or NO RESPONSE
NATO (Liberal (L)) Move from 400m towards 25m until you reach a HIT
Move a “4 block window” from 400m towards 25m until “3 out of 4”
“3 out of 4” Hits (3) HITS have been achieved then select the furthest HIT achieved; if no
“3 out of 4” exists, go with Conservative

The “NATO” rule is labeled liberal because it generates the longest detection distances. The
“Conservative” rule resulted in the shortest detection distances, and “3 out of 4” can be described as an

10
intermediate rule. Using these three rules, up to three different detection values were obtained for a given
observer. Application of the three rules to each line of observer data is also illustrated in Figure 11 by the
use of “L” for “Liberal”, “C” for “Conservative” and “3” for “3 out of 4”. The distance in meters for
each column is noted across the top, while subject identification for each row is noted on the left hand
side of the table. A row is one line of data for one observer’s detection of a given pattern in one series.
To determine the Conservative detection distance for Row A, start from 25 m, and move to the right. The
last hit (i.e., green block) before the first miss or no response (i.e., red or yellow block) is the conservative
detection distance, 150 m. To determine the Liberal detection distance for the same subject, start at 400
m and move to the left. The first hit, which is at 250 m, is the Liberal detection distance. To determine
the “3 of 4” detection distance, move from right to left using a sliding group of four blocks. The first
group that has three out of four blocks marked as hits, determines the data area of use. The furthest hit in
the block of four is the detection distance for the 3 of 4 rule. For Row A, 250 m is the 3 of 4 detection
distance. Miss and no response observer data were treated equally for the analysis.

25 50 75 100 125 150 175 200 250 300 400


A C L3
B C3 L
C CL3
D C 3 L
E C L3
F C3 L
G C3 L
H CL3
I C L3
J C L3
K C L3
L C3 L
M C3 L

< - Hit
<- Miss
<- No Response

Sample Observer Data


Figure 11

3.1 Data Adjustments

During the photosimulation evaluation, observers made detections at pre-defined intervals (400, 300, 250,
etc.) without the opportunity for detections between intervals (399 through 301, 299 through 251, etc.).
Therefore, the observer’s detection value is biased towards the lower distance (i.e., the range 399 to 300
is actually represented by 300). It is assumed the actual detection would have taken place somewhere
between the actual detection distance and the next highest detection distance. To minimize this bias, a
uniform-spacing method is used. The total number of detections in an interval are equally spaced
between the distance the detection took place and the next furthest distance value. For example, if four

11
observers detected Pattern A at 50 m, the four detection values would be evenly spaced between 50 and
75 m (i.e., 55 m, 60 m, 65 m and 70 m). The formula below is used to calculate uniform-spacing:

Spacing:
(Ra – Rb)/(n+1)

Ra = range where “n” detections took place


Rb = next longer range

4.0 Results
4.1 Medians by Rule

The results, when viewed by different rules (Figure 12), show little difference between the Liberal and 3
out of 4 rules. The Conservative rule typically generates the shortest detection distance, while the Liberal
and 3 out of 4 generate the longest detection distance. The Conservative rule ignores data with unusual
patterns. If there are hits in between misses/no responses, the Conservative rule will ignore those hits and
only use the hit at the shortest distance. The results presented in this report are based on the 3 out of 4
rule.

(b) (2)

Desert

Urban

Woodland

Medians by Rule
Figure 12

12
4.2 Medians by Pattern / Environment

Figure 13 shows the individual pattern detection median for 50% probability (R50) of detection by
environment. For ease of reading, the graphs are sorted by the top bar graph, desert. The UCP pattern is
designated with a red bar for easy reference; desert patterns are tan, commercial patterns are grey, and
woodland patterns are green. The patterns dropped from the final analysis as discussed in Section 2.4
appear as white bars. The five dropped patterns are not discussed in the remainder of this report, but data
on each pattern are available for additional analysis.

As expected, the desert patterns were detected at shorter distances to the target in the desert environments
while the woodland patterns were detected at the shorter distances in the woodland environments. In the
urban environment all targets were detected at relatively longer distances.

(b) (2)
Meters

Desert
Meters

Urban
Meters

Woodland

Pattern R50 by Environment


Figure 13

13
The range of detection medians for each environment is stated in Table 6. The range of detection
medians is narrower in urban environments and wider in woodland environments.

Range of Medians by
Environment
Table 6
Data range
Desert 115 – 400
Urban 130 – 400
Woodland 36 – 400

Figure 14 is a visual depiction of the ranges presented in Table 6 but broken out by the nine scenes. The
maximum detection distances in the three urban scenes are all out to 400 m. Patterns were detected at
further distances in the woodland sun scene because the lighter patterns were easily detected in this scene.
When direct light was decreased or eliminated in the woodland environment, as in the shade and partial
sun scenes, detection distances decreased. The desert rock scene was difficult for the observers to detect
the targets. The rocks provided enough clutter in the scene to make immediate detection difficult. The
desert sage scene generated farther detection distances. This may be due to the target being taller than the
sage surrounding him; detection in this scene may have been due to detection of target shape, more than
detection of target pattern. The desert sand scene was a low clutter environment where the lighter
uniforms camouflaged better with the immediate surroundings.

(b) (2)

Desert

Urban

Woodland

Range of Data for Nine Scenes


Figure 14

14
4.3 Ranking

The patterns were ranked according to median detection distance and are presented in Table 7. If medians
were equal, then the mean was used for ranking. As expected, the desert patterns and woodland patterns
performed the best (i.e., lower medians) in their respective environments. By comparison, the desert
patterns had poor performance (i.e., higher medians) in the woodland scenes, as did the woodland patterns
in the desert scenes. It is noted that a pattern that ranked best in one scene (e.g., desert rock) did not
necessarily rank as well in the two remaining scenes.

Ranking of Patterns by Scene


Table 7
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Partial Shade
Desert Brush MultiCam® Desert British Desert MARPAT Desert Brush Desert Brush China China China
Best Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush MultiCam® Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq MultiCam® French Iraq Mossy Oak
Bulldog Desert MARPAT MultiCam® China UCP UCP Bulldog Syria French
MultiCam® Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear MultiCam® Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak MultiCam® Iraq Desert British Desert Brush MultiCam®
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French MultiCam® Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq French Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
Worst China UCP Iraq Woodland MARPAT French Woodland MARPAT UCP Desert British UCP

Ranking of Patterns by Environment


Table 8
Desert Urban Woodland
Desert British Desert MARPAT China
Syria Desert Brush Woodland MARPAT
Best
Desert MARPAT Desert British Mossy Oak
Desert Brush MultiCam® French
Bulldog Syria Iraq
MultiCam® Natural Gear MultiCam®
Natural Gear Iraq Syria
Mossy Oak UCP Bulldog
UCP Bulldog Natural Gear
French China Desert Brush
Woodland MARPAT Mossy Oak Desert MARPAT
Worst
China Woodland MARPAT UCP
Iraq French Desert British

In order to rank the median detection distance by environment, the data were collapsed over the three
scenes within an environment (Table 8). As before, the data shows desert patterns performing well in the
desert environment and the woodland patterns performing well in the woodland environment. Of the
commercial patterns evaluated, the Mossy Oak pattern performed much like the woodland patterns,

15
possibly due to the dark colors used. Given these environment-specific results, five patterns consistently
performed better than UCP – Syria, Desert MARPAT, Desert Brush, MultiCam®, and Natural Gear.
Although the performance of the Natural Gear pattern was consistently better than UCP, it is not a viable
pattern to consider for possible near-term military use because it does not have military near-infrared
properties and the Government has no rights to the pattern. Natural Gear was also the lowest performer
of this group of five patterns. Therefore, further discussion is focused on the remaining four uniforms –
Desert MARPAT, Desert Brush, MultiCam® and Syria. Syria, although a foreign uniform and not
practical for U.S. military use, yielded very favorable results in the environments tested and may be
useful in future research on pattern / color effectiveness in multiple environments.

4.4 Results By Pattern

Statistically significant results (p < .05 based on Mann-Whitney test of significance) for the four patterns
that performed consistently better than the UCP are described below. The tables in this section present
the same pattern ranks as in Table 7, but they also indicate performance that is statistically better than,
equal to and worse than a given pattern, which is highlighted by a bold outline. A highlighted pattern has
the possibility of being significantly different from 108 data points (i.e., 12 remaining patterns x 9
scenes). If a pattern appears in a green, grey or tan block, then it has a median detection distance that is
significantly worse than the highlighted pattern. If a pattern appears in a white block, then it has a
median detection distance that is significantly better than the highlighted pattern. If a pattern appears in a
white block but its name is crossed off, then it has a median detection distance that is not significantly
different from the highlighted pattern.

4.4.1 UCP: The UCP (Table 9) performed significantly better than 11 of the 108 data points – five
patterns in the desert scenes and six patterns in the urban scenes. The performance of the UCP was
significantly better in certain desert and urban scenes compared to the darker woodland and commercial
patterns. It did not perform significantly better than any pattern in either the woodland environment or the
desert sage scene. Sixty-two data points were ranked as significantly better than the UCP – 19 patterns in
desert scenes, 10 patterns in urban scenes, and 33 patterns in woodland scenes.

Statistically Significant Results Relative to the UCP


Table 9
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Partial Shade
Desert Brush MultiCam® Desert British Desert MARPAT Desert Brush Desert Brush China China China
Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush MultiCam® Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq MultiCam® French Iraq Mossy Oak
Bulldog Desert MARPAT MultiCam® China UCP UCP Bulldog Syria French
MultiCam® Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear MultiCam® Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak MultiCam® Iraq Desert British Desert Brush MultiCam®
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French MultiCam® Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq French Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
China UCP Iraq Woodland MARPAT French Woodland MARPAT UCP Desert British UCP
Str ke hrough Pattern median is not significantly different than highlighted pattern
Italics Pattern median is significantly better than highlighted pattern

Table Key Red


Green
UCP
Woodland Pattern significantly worse than highlighted pattern
Grey Commercial Pattern significantly worse than highlighted pattern
Lt Brown Desert Pattern significantly worse than highlighted pattern

4.4.2 Desert Brush: The Desert Brush (Table 10) pattern is an Army developmental pattern. In this
evaluation, it performed significantly better than 53 of the 108 data points – 25 patterns in desert scenes,

16
21 patterns in urban scenes, and seven patterns in woodland scenes. Desert Brush performed significantly
better than UCP in six of the nine scenes. Even though 30 patterns performed significantly better than
Desert Brush, only one was in a desert scene, while the bulk (i.e., 25 data points) were associated with the
woodland environment. Its desert and urban performances were quite favorable.

Statistically Significant Results Relative to the Desert Brush


Table 10
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Partial Shade
Desert Brush MultiCam® Desert British Desert MARPAT Desert Brush Desert Brush China China China
Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush MultiCam® Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq MultiCam® French Iraq Mossy Oak
Bulldog Desert MARPAT MultiCam® China UCP UCP Bulldog Syria French
MultiCam® Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear MultiCam® Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak MultiCam® Iraq Desert British Desert Brush MultiCam®
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French MultiCam® Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq Iraq Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
China UCP French Woodland MARPAT French Woodland MARPAT UCP Desert British UCP
Red Border Highlighted Pattern of Interest
Strike-Through Pattern median is not significantly different than highlighted pattern
Italics

Table Key
Pattern median is significantly better than highlighted pattern
Red UCP
Green Woodland Pattern significantly worse than highlighted pattern
Grey Commercial Pattern significantly worse than highlighted pattern
Lt Brown Desert Pattern significantly worse than highlighted pattern

4.4.3 Desert MARPAT: This pattern is the present desert combat pattern used by the United States
Marine Corps (USMC) (Table 11). It ranked significantly higher than 45 of the 108 data points – 22
patterns in desert scenes, 18 patterns in urban scenes and five patterns in woodland scenes. It ranked best
in one of the nine scenes; in eight of the nine scenes it ranked significantly better than the UCP. The one
scene where it ranked lower (urban small building) than the UCP, there is no significant difference
between their rankings. A total of 29 data points were significantly higher than the Desert MARPAT,
with 22 occurring in the woodland environments.

Statistically Significant Results Relative to the Desert MARPAT


Table 11
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Partial Shade
Desert Brush MultiCam® Desert British Desert MARPAT Desert Brush Desert Brush China China China
Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush MultiCam® Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq MultiCam® French Iraq Mossy Oak
Bulldog Desert MARPAT MultiCam® China UCP UCP Bulldog Syria French
MultiCam® Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear MultiCam® Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak MultiCam® Iraq Desert British Desert Brush MultiCam®
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French MultiCam® Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq French Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
China UCP Iraq Woodland MARPAT French Woodland MARPAT UCP Desert British UCP
Red Border Highlighted Pattern of Interest
Strike-Through Pattern median is not significantly different than highlighted pattern
Italics

Table Key
Pattern median is significantly better than highlighted pattern
Red UCP
Green Woodland Pattern significantly worse than highlighted pattern
Grey Commercial Pattern significantly worse than highlighted pattern
Lt Brown Desert Pattern significantly worse than highlighted pattern

4.4.4 MultiCam®: This pattern (Table 12) is an Army developmental and commercially available
pattern. MultiCam® performed significantly better than 44 of the data points – 17 patterns in desert
scenes, 15 patterns in urban scenes and 12 patterns in woodland scenes. The consistency of the pattern
performance across all three environments indicates a potential to be a viable universal pattern. It
performed significantly better than UCP in seven of the nine scenes. With regard to the other two scenes

17
(i.e., desert sand and urban small building), there was no significant difference between MultiCam®’s
and UCP’s performances. Thirty-four data points are significantly better than MultiCam® – 9 patterns in
desert scenes, 5 patterns in urban scenes, and 20 patterns in woodland scenes.

Statistically Significant Results Relative to the MultiCam®


Table 12
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Partial Shade
Desert Brush MultiCam® Desert British Desert MARPAT Desert Brush Desert Brush China China China
Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush MultiCam® Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq MultiCam® French Iraq Mossy Oak
Bulldog Desert MARPAT MultiCam® China UCP UCP Bulldog Syria French
MultiCam® Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear MultiCam® Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak MultiCam® Iraq Desert British Desert Brush MultiCam®
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French MultiCam® Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq French Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
China UCP Iraq Woodland MARPAT French Woodland MARPAT UCP Desert British UCP
Red Border Highlighted Pattern of Interest
Strike-Through Pattern median is not significantly different than highlighted pattern
Italics

Table Key
Pattern median is significantly better than highlighted pattern
Red UCP
Green Woodland Pattern significantly worse than highlighted pattern
Grey Commercial Pattern significantly worse than highlighted pattern
Lt Brown Desert Pattern significantly worse than highlighted pattern

4.4.5 Syria: The Syrian pattern (Table 13) performed well in seven of the nine scenes. It performed
significantly better than 53 of the 108 data points – 22 patterns in desert scenes, 17 patterns in urban
scenes, and 14 patterns in woodland scenes. The consistency of the pattern performance across all three
environments indicates a potential to be a viable universal pattern. Only 22 data points performed
significantly better than Syria, and over half (i.e., 13 data points) were associated with the woodland
environment. Syria performed significantly better than UCP in six out of nine scenes. As for the
remaining three scenes, there was no significant difference between Syria’s and UCP’s performances.

Statistically Significant Results Relative to Syria


Table 13
Desert Urban Woodland
Rock Sage Sand Door Small Bldg Corner Bright Sun Shade Patrial Shade
Desert Brush Multicam Desert British Desert MARPAT Desert Brush Desert Brush China China China
Desert British Natural Gear Syria Natural Gear Woodland MARPAT Desert MARPAT Mossy Oak Woodland MARPAT Syria
Syria Syria Desert MARPAT Desert British Syria Desert British Iraq French Iraq
Desert MARPAT Desert Brush Desert Brush Multicam Bulldog Syria Woodland MARPAT Mossy Oak Woodland MARPAT
Natural Gear Desert British Bulldog Iraq Iraq Multicam French Iraq Mossy Oak
Bulldog Desert MARPAT Multicam China UCP UCP Bulldog Syria French
Multicam Mossy Oak China Desert Brush Desert MARPAT Bulldog Natural Gear Multicam Natural Gear
Woodland MARPAT Bulldog Natural Gear Mossy Oak Multicam Iraq Desert British Desert Brush Multicam
French China UCP French Desert British Natural Gear Desert MARPAT Desert MARPAT Desert Brush
UCP French Mossy Oak UCP China French Multicam Bulldog Bulldog
Iraq Woodland MARPAT Woodland MARPAT Syria Natural Gear China Syria Natural Gear Desert MARPAT
Mossy Oak Iraq Iraq Bulldog Mossy Oak Mossy Oak Desert Brush UCP Desert British
China UCP French Woodland MARPAT French Woodland MARPAT UCP Desert British UCP
Red Border Highlighted Pattern of Interest
St ike-Th ough Pattern median is not significantly different than highlighted pattern
Italics

Table Key
Pattern median is significantly better than highlighted pattern
Red UCP
Green Woodland Pattern significantly worse than highlighted pattern
Grey Commercial Pattern significantly worse than highlighted pattern
Lt Brown Desert Pattern significantly worse than highlighted pattern

4.4.6 Performance: Figure 15 shows the detection range improvement in meters and percent, over the
UCP, of the four patterns of interest (e.g. MultiCam®, Desert Brush, Desert MARPAT, and Syria) in
each of the three environments. It also shows the best performing patterns in each of the environments.
Woodland MARPAT is shown as a good U.S. woodland pattern. However, like most woodland patterns,
it was relatively ineffective as a universal pattern due to its poor performance in desert and urban
environments.

18
In the desert environment, the MultiCam® pattern median detection point was 16% (b) (2) better than
(b) (2)
that for the UCP, while Soldiers wearing the Desert Brush and Desert MARPAT patterns can get 28%
to 34% (b) (2) closer, respectively, to the observer than those wearing the UCP. The Desert British
pattern had the greatest improvement (37% / (b) (2) in the desert environment.

In the urban environment, MultiCam® provided a 16% (b) (2) improvement in median detection distance
compared with UCP. Again, Desert Brush and Desert MARPAT performed slightly better than
MultiCam®, showing 22% (b) (2) and 24% (b) (2) respectively, improvement over UCP. Desert
MARPAT was the best performer in the urban environment.

In the woodland environment, MultiCam® provided a 36% (b) (2) improvement in median detection
distance, while Desert MARPAT and Desert Brush provided less advantage, with 16% (b) (2) and 20%
(b) (2) improvement, respectively. The darker woodland patterns performed much better, with the China
pattern providing a 77% (b) (2) improvement in median detection distance over UCP. Data on all of the
woodland patterns are available for further analysis in this specific environment.

Figure 15 shows the trade-off made when looking for a pattern that performs well in all environments.
Clearly, desert patterns perform best in desert environments, while woodland patterns perform best in
woodland environments. In all environments, the four patterns of interest provided an improvement in
detection as compared to UCP. The range of improvement over the three environments was 10-36%.

(b) (2)

Patterns Of Interest Best Performers by Environment


UCP UCP DM Desert MARPAT
Desert DE Desert British
MC MultiCam DB Desert Brush
Urban DM Desert MARPAT
SY Syria WM Woodland MARPAT
Woodland CH China

Differences in Median Detection Distances Relative to the UCP


Figure 15

19
5.0 Discussion

Overall, in the woodland environment, the lighter colored patterns were detected at further distances than
the darker patterns. The opposite was found in the desert and urban conditions. These data confirm what
is intuitive regarding the performance of environment-specific patterns: woodland patterns perform best
in woodland environments, and desert patterns perform best in desert environments.

Although the detection distances for the UCP did not fluctuate tremendously between environments, the
performance of other patterns indicated that the visual detection of a Soldier can be enhanced
significantly in all environments. The four evaluated patterns of interest (MultiCam®, Desert MARPAT,
Desert Brush, Syria) improved the Soldier’s visual detectability by decreasing the detection distance by a
minimum of 16% in the desert and woodland environments as compared to the target wearing UCP.

The Desert MARPAT is a USMC-specific pattern with Government rights. Visual and near infrared
(NIR) standards are in place for this pattern, which is in full-scale production. It performed very well in
the desert/urban environments, with marginal performance in the woodland. This pattern did have
significant performance improvement over the UCP in eight of the nine scenes evaluated.

The Desert Brush pattern is a Government-owned pattern. It has visual and NIR performance
characteristics, but lacks standards for visual, NIR and shortwave infrared (SWIR). To date, it has been
printed on nyco and Cordura® fabrics by one printing company. Approximately six months lead time
would be needed to get this pattern ready for full-scale production. Desert Brush performed very much
on par with Desert MARPAT, performing well in the desert/urban environments and marginally in the
woodland. Desert Brush performed better than the UCP in all environments and significantly better in six
of the nine scenes evaluated.

The MultiCam® pattern was developed under a U.S. Army development contract and is commercially
available. Visual, NIR and SWIR standards have been established on three fabric substrates (i.e., nyco
ripstop, 500 denier Cordura®, and Defender™ M - a flame resistant material) printed by a single
company. The authors are aware of only one other company that has printed MultiCam® with good
visual and NIR properties. MultiCam®’s performance in the desert and urban environments was not as
good as the Desert MARPAT and Desert Brush patterns; however, it was significantly better than both
patterns in the woodland environment. MultiCam® performed significantly better than the UCP in seven
of the nine scenes evaluated.

The Syrian pattern, while not a viable pattern of interest for U.S. Army use, exhibited good performance
as a universal pattern. Its performance in both the desert and woodland environments is something to be
studied if pattern development is pursued. It performed on par with the best desert pattern and on par
with the best universal pattern, i.e., MultiCam®.
As stated earlier, neither the Woodland nor the Desert Battle Dress Uniforms (BDU) were included in this
evaluation. At the time patterns were selected for evaluation, neither BDU pattern was being used by the
U.S. Army.

20
6.0 Conclusions and Recommendations

There are many alternatives to consider with regard to camouflage for the U.S. Army. Below are some
recommendations on just a few possible courses of action:

1. The data clearly show that environment-specific patterns provide the best camouflage, i.e., lowest
probability of detection, in their respective environments. These data clearly indicate that two pattern
types, woodland and desert/urban, will provide the best camouflage to the Soldier with missions in
these specific environments.

a. If needed, further analysis of this data set can be done, with specific regard to the woodland
environment, to assist with pattern selection or optimization.
b. Urban environments pose a particular challenge in terms of camouflage development due to the
diversity within an urban area and between urban areas.

2. If Army leadership desires, for any number of reasons, to maintain a single, multi-environment
camouflage pattern for combat missions, then one must first consider all possible environments that a
Soldier can encounter during a mission set. For instance, in present day theaters, Soldiers can
manuever from desert mountainous terrain to oasis to urban terrain during a single mission.
MultiCam® provides a readily available alternative with good overall performance across all three
environments.

a. It provides a significant reduction in target detectability in all three environments as compared to


the UCP. MultiCam® performed better in the woodland environment than the Desert MARPAT
and Desert Brush patterns, while those two patterns performed better in the desert environment
than MultiCam®.
b. Specific woodland environment missions may still need to be supplemented with a woodland
pattern.

3. Feedback from post combat surveys has indicated Soldier dissatisfaction with the UCP colors, but not
with the digital pattern itself. Development of a color-modified UCP would require additional testing
to confirm detection and blending effectiveness, and adoption would have the same financial impact
discussed in item 4 below.

4. Changing the Army camouflage pattern is a multi-billion dollar issue as it requires a change not only
to the uniform, but also to all the clothing and individual equipment items. Options must be carefully
weighed regarding this issue. Options include, but are not limited to:

a. Keeping the UCP as a garrison uniform, while supplementing combat missions with either an
improved multi-environment pattern, such as MultiCam®, or environment-specific patterns.

b. For most personal protective equipment (PPE) and individual equipment items, adopting a solid
color that works well with all combat uniform patterns. This is the strategy the USMC has used
with their Desert and Woodland MARPAT uniforms and solid coyote-colored PPE and load

21
carriage gear. Potential survivability risks associated with this option have not been characterized
by the Army.

5. While development of a new and improved multi-environment pattern is an option, the authors of this
report do not believe that significant visual detection performance advances are achievable across all
environments over the 10-36% improvement already demonstrated by the four patterns of interest.
However, limited incremental improvements may be achievable using current textile technology.

a. Other aspects that may affect camouflage performance, such as color count, pattern size, color
contrast and printing effects, may be considerations in a future pattern development, as these
factors potentially impact cost, producibility and durability.

6. Clear requirements are the key to successful implementation of any future camouflage pattern(s).
This data set, along with additional studies or specific surveys, may be useful in clearly defining the
camouflage pattern requirements.

7. Areas of consideration for future camouflage development include the analysis of mission sets within
and across environments and the study of camouflage during the Soldier’s movement within
operational scenarios verses the static conditions of this study. Also, a thorough review of industry’s
research and development efforts may introduce novel techniques and advances for military
camouflage. For instance, the impact of new industrial capabilities, such as high speed inkjet printing,
can be investigated. These capabilities may provide a faster industry response to enable rapidly
deployable camouflage specific to a theater of interest.

09 / 021L

22
7.0 References

1. Computerized Visual Camouflage Evaluation; (b) (6)(b) (6)(b) (6)(b) (6)(b) (6)
(b) (6)(b) (6)(b) (6)(b) (6)(b) (6) April 2008. In Press.

2. 2007 Parallel Meetings of the Military Sensing Symposia (MSS) Specialty Groups on
Passive Sensors, Camouflage, Concealment and Deception; Detectors; and Materials, Volume 1,
March 2007 (MSS-PSCC-07-0046).

3. Guidelines for Camouflage Assessment Using Observers, NATO RTO-AG-SCI-095, Oct


2006.

4. Enhancement of Camouflage Assessment Techniques, RTO-TM-SCI-095, NATO RTO


Systems Concepts and Integration Panel Task Group SCI-095 final report, December 2006.

5. Signature Measurements Guidebook, Glenn Rogers, US Army Aberdeen Proving Grounds,


Sept 2007.

6. SPSS for Windows, Rel. 11.0.1. 2001. Chicago: SPSS Inc.

23
(This page has intentionally been left blank.)

24
Appendix A.
Observer Instructions

The ultimate purpose of this work is to improve the effectiveness of land forces through the improvement
of camouflage. You will be asked to act as an observer looking for a camouflaged person (target) in a
series of pictures. The basic idea of this test is to determine the range at which you detect the target.

You will view 9 different series of pictures. Each series consists of 11 pictures. The first picture in the
series is at a far distance from the target; the following pictures gradually bring you closer to the target.
Each picture is shown for a maximum of 14 s.

While viewing the pictures, if you detect a target, click on it using the mouse. You will not receive any
positive or negative feedback when you click on the target. Once you click on a target, you can then use
the next button, located in the lower right hand corner to move quicker through the series. You will
continue to view all 11 slides in the series so please click on the target in each slide until the end of the
series.

The nine series that you are going to view consist of three woodland, three urban and three desert.

You are looking for a camouflage person, not a vehicle or any type of equipment.

You are looking for ONLY on person in each series.

It is a test of the camouflage patterns, not a test of your personal ability.

Your name will not be recorded on the results.

Do not discuss this test with others.

25
(This page has intentionally been left blank.)

26
Appendix B.
Random Orders for Each Pattern

Run
Number Uniform Series 1 Series 2 Series 3 Series 4 Series 5 Series 6 Series 7 Series 8 Series 9
1 FFW D2-2 D3-2 W3-3 W2-1 D1-3 U2-1 W1-3 U3-1 U4-2
2 ACU D3-2 W1-2 W3-3 W2-1 D1-3 U2-1 W1-3 U3-1 U4-2
3 D. British U2-2 W3-2 D1-2 U4-1 D2-1 W1-1 U3-3 W2-3 D3-3
4 Bulldog W2-1 U3-3 D2-2 W1-2 U2-2 D3-3 U4-1 W3-3 D1-3
5 D. MARPAT D1-2 W3-2 W1-3 U4-1 D3-3 W2-1 U2-3 D2-1 U3-2
6 W. MARPAT W2-2 U3-2 D2-3 W3-1 U4-3 D3-1 D1-3 U2-1 W1-2
7 D. Brush U4-2 D2-2 D1-3 W1-1 U3-3 W3-1 D3-3 W2-1 U2-2
8 China W3-2 U2-2 W2-3 D2-1 W1-3 U3-1 U4-3 D3-1 D1-2
9 French U4-3 D1-1 W1-2 D2-2 U2-1 W3-3 U3-2 D3-3 W2-1
10 Iraq U2-2 D3-2 D2-3 D1-1 W2-3 W1-1 U4-3 U3-1 W3-2
11 Mossy Oak D3-2 U3-2 W2-3 U4-1 W3-2 D1-1 W1-3 U2-1 D2-2
12 N. Korea U3-2 U2-2 W3-3 W1-1 D2-3 U4-1 D1-3 D3-1 W2-2
13 Sweden W3-2 D2-2 U4-3 U3-1 W1-3 D3-1 U2-3 W2-1 D1-2
14 Syria U4-2 W1-2 D3-3 W3-1 U3-3 W2-1 D2-3 D1-1 U2-2
15 W. British D1-2 U3-3 W1-2 U2-1 D2-1 W2-1 U4-2 D3-3 W3-3
16 FFW (Cloud) D1-2 U4-2 U1-3 W2-1 D3-3 D2-1 W3-3 W1-1 U3-2
17 ACU (Cloud) W2-2 W3-2 D1-3 D3-1 U4-3 U1-1 U3-3 D2-1 W1-2
18 N. Gear D2-2 W2-2 W1-3 U2-1 D1-3 U3-1 D3-3 W3-1 U4-2
19 FFW (Cloud) U3-2 W1-2 D2-3 D1-1 W2-1 U4-3 D3-1 W3-1 U2-3
20 ACU (Cloud) W2-2 U4-3 D2-3 W3-1 U3-2 D1-3 W1-1 D3-3 U2-1
21 FFW W2-2 D3-3 U3-3 D2-1 U2-2 W1-1 D1-2 W3-2 U4-1
22 ACU U3-3 W1-1 W3-2 D2-1 U2-2 D1-2 W2-1 D3-1 U4-2
23 Spec4W W3-1 U4-1 U3-2 W1-3 W2-2 U2-1
24 Bulldog W1-1 U4-2 D3-2 W2-3 D2-1 W3-2 U3-1 D1-3 U2-3
25 D. MARPAT U2-1 W1-2 D3-2 W3-1 D1-3 U4-3 D2-2 W2-3 U3-1
26 W. MARPAT W3-2 D1-1 U3-3 W2-1 U4-2 W1-3 U2-2 D3-3 D2-1
27 D. Brush W1-2 D2-1 U4-3 W3-2 W2-3 U3-2 D1-1 U2-3 D3-1
28 Mossy Oak D2-3 U4-2 D1-2 U2-3 W3-1 U3-2 D3-1 W1-1 W2-3
29 Syria W2-3 D3-1 U3-2 W1-3 D1-2 U4-1 D2-1 U2-3 W3-2
30 N. Gear D1-2 W3-3 U2-2 U4-1 W2-3 D2-1 W1-2 U3-3 D3-1
31 Spec4U U4-2 W2-1 W3-2 U2-3 W1-3 U3-1
32 China W1-2 D3-3 D2-2 U2-1 W3-3 D1-1 U4-2 W2-1 U3-3
33 N/A
34 French W3-2 D2-1 U3-3 W2-2 D1-2 U4-1 D3-3 U2-2 W1-3
35 Iraq U3-3 W2-1 D2-2 U2-1 W3-3 U4-1 D1-3 W1-2 D3-1
36 N. Korea D1-2 W1-2 U2-1 D3-3 W3-1 U3-3 D2-1 W2-3 U4-2
37 Sweden D3-2 U3-3 W2-2 D2-1 W3-3 U4-2 D1-3 W1-1 U2-1
38 W. British W3-1 D3-2 D1-1 W2-2 U2-2 U4-3 D2-3 W1-3 U3-1
39 Spec4W U3-1 W3-2 W1-1 U4-2 U2-3 W2-3
40 D. British D1-3 W1-2 U4-3 W2-1 D3-2 U2-1 W3-3 D2-1 U3-2
41 Spec4U W3-1 U2-2 W1-2 U3-3 W2-3 U4-1

27

You might also like